Agile, API-based systems are helping both MuleSoft customers and Salesforce react to the new business reality of increased remote work and rapid digital transformation.
In conjunction with Salesforce’s MuleSoft CTO Uri Sarid for out podcast. Sarid discussed a variety of topics, including the company’s continued efforts to help companies build API-based infrastructures and capabilities, how these API-based systems are helping both customers and Salesforce adapt to the new normal of the COVID-19, and what’s on the horizon for MuleSoft, such as a new feature they’ll be releasing later this year called API Federation. The following is a transcript of our interview, edited for readability.developer event, which was completely virtual this year, I had a chance to speak with
Bill Detwiler: So, when you and I spoke last at Dreamforce last year, we had a great conversation about how MuleSoft, and how you, really see the future for APIs as being plug and play. As trying to make them easy to create so that it can enable organizations, companies, to really kind of bring products and services to their customers much more rapidly than they had in the past. And so I wanted to talk to you now about what work kind of MuleSoft is doing around that, and the tools that you’re developing to help people, to help companies, as they respond to the new normal that we find ourselves in with the COVID-19 pandemic, and people trying to figure out how to return to work and do it safely.
Uri Sarid: Yeah, absolutely. It’s interesting, when we talked last at Dreamforce, I think we set a pretty wide and ambitious agenda for the world of technology and information and so on. Based on APIs, it’s basically a starting point to show that once you expose your capabilities as productized interfaces, lots of people can go in and compose those together and benefit from those capabilities, and that allows, as you said, everyone to go faster.
Uri Sarid: Usually when we talk about these capabilities we come up with use cases. If you wanted to do better e-commerce, you could plug and play these APIs in the following way and so on. If I were at that point back at Dreamforce to tell you, “Imagine a use case where a global pandemic keeps everyone at home and makes everything remote, and now you have to respond to it in a matter of sometimes days or weeks,” you would’ve said, “Yeah, but let’s talk about real use cases. Come on. Enough talking about some future.” Right?
Uri Sarid: Yet here we are living in the mother of all use cases, and so I think that topic has become very, very relevant. What’s perhaps not surprising is that we as MuleSoft are not explicitly saying, “Well, you know what? Now we’re going to pivot 45 degrees in one direction.” One, because like most companies, pivoting radically and kind of throwing out everything you did before is simply not feasible. In our case, it wasn’t necessary. And to a large extent it’s because the customers who have built these kind of API best agile infrastructures are looking to do more of that, the customers who haven’t are scrambling to get there, and the ones who have built these kinds of API based infrastructures or capabilities have been able to actually pivot themselves very, very rapidly and adjust to the new reality.
Uri Sarid: So what we’re doing now is actually basically just doubling down on that. How can we make it happen even faster, and in many cases, how do we consume those capabilities ourselves? I’d love to talk to you more about that, and be of service to our stakeholders. Actually, not just our customers, but our communities. Right? Literally, basically, how do we attach to changing the outcome of the pandemic and the economic outcome of that as rapidly as possible?
SEE: Hiring kit: Salesforce Developer (TechRepublic Premium)
Bringing data and resources together quickly in response to COVID-19
Bill Detwiler: Yeah, let’s talk about that a lot. Because I know it’s one of the things that we’ve talked about before, that MuleSoft does is you use the tools internally that you develop for customers. I know that one of the things you’ve been working on is how to integrate the APIs to help with what Salesforce has been doing on the Work.com platform and on the Crisis Response Developer Portal to bring the data together, to bring all the platforms together in a way that they can be used by a lot of people really kind of quickly. Talk about how that came about. Talk about MuleSoft and the APIs role in allowing that to happen.
Uri Sarid: Yeah, I’ll be very realistic and practical. We got an inkling that this was going to go in a very intense direction pretty early on. I think we were one of the earliest companies here to switch to a working from home mode, and working with leadership it became very clear that the world was going to change very, very rapidly.
What we saw in literally the first few weeks of the pandemic hitting us is that our stakeholders, meaning the public, the communities that we live in, and of course our own employees and our customers, were going to have to make a lot of very rapid decisions, and they were doing it based on data that was a very fragmented. Even the newspapers started to track the infection of the pandemic, and where actual cases were happening and so on and publishing that in whatever format made sense to them, and we saw that the consumers of that, whether it’s mobile application builders or website builders or decision makers, were all going to have to reinvent the ability to digest that data and put it together in a meaningful way.
So we said this is something that we can actually do. We’re not doing it for business purposes, we’re doing it in order to be part of the community that’s going to change the outcome of the pandemic. We ended up deciding to build a COVID-19 data platform, whose purpose would be to take all of these myriad data sources, put them in front of some experts so we could quickly curate that data, take that data and normalize it. Put it in standard formats, persist it and offer it back out to anybody who wanted it for free. Whether it’s through an API that a mobile developer can go and consume that data, or whether it’s through a Tableau visualization where decision makers can slice and dice and say, “Hey, this is the right thing for me to do in a particular location based upon real data, like where is the virus and how prepared am I and how much critical resources do I actually have?”
We used our own technology, we used MuleSoft technology, Tableau technology. We had great partners that we ended up working with, and that orientation towards making data based decisions we thought was our contribution to that. We also, as you mentioned, took the opportunity to take all of those APIs and all of the patterns that we learned and publish it into a developer resource portal that is particularly oriented towards this crisis.
In fact, that’s not just for the pandemic data. We also have healthcare integrations, and we’ve created a healthcare accelerator to allow hospitals to basically respond to the triage and triage their patients. God forbid if we have a lot more spikes of this pandemic, as may well happen, hospitals will be better prepared to be able to serve their needs. And again, those all went into that same developer portal.
What we also saw, and again this is something that you mentioned, that the next phase would be that businesses would have to open up or else the economic outcome of this thing is going to be incredibly disastrous as well. And so, how can they open up responsibly? In the past, we used to talk about how do you innovate rapidly and have security? Well now we’re talking about how do you open up rapidly and actually do it in a responsible way? That’s where the Work.com initiative comes in. It’s a wrapper to take a bunch of capabilities that Salesforce has, including the data platform, and put it together in a way that allows our customers to reliably open up some locations, close other locations, and make those decisions based on real data.
Bill Detwiler: Do think it’s possible to bring all those disparate data sources together into a usable portal like that, into a usable platform, without something like MuleSoft’s integration layer? Without the API technology that we have it doesn’t seem that it’s actually possible, or especially not to do as quickly, to bring these different sources of information together, to allow decision makers to take action based on good information.
Uri Sarid: Yeah. It’s always possible to do anything with software, but the key, as you said, is can you do it rapidly, can you do it efficiently? And MuleSoft was a big part of that. Frankly, we were not the only part of that. Right? So we’re really good at attaching to existing sources, transforming them in a very robust way, putting them somewhere. We ended up creating a data link. Actually to be specific, a data store in a data warehouse, where the information is already normalized and structured. It’s not unstructured data. And then serving it back out as APIs, or transporting it to the Tableau data hub and making sure that it’s available for visualization.
But we’re not the only technology provider at Salesforce, and the rest of Salesforce was also very important. So as we talked about last year, we have a lot of expertise around tooling and modeling, and Salesforce has a lot of expertise around modeling across lots of areas. CIM, the Cloud Information Model, is part of that. There are FHIR and other healthcare APIs that have normalized the models, and so we’re leveraging the Salesforce expertise around modeling some industry standard models, we’re leveraging Tableau technology to prep some of that data, and so we really came together in all of this and were able to pull off the data platform in a very, very short amount of time.
Bill Detwiler: Talk a little bit about that coming together. It’s hard enough to bring different companies together, different agencies together, public/private partnerships together, but it can be difficult even within a company to bring different BU’s together, especially if like Salesforce, a lot of those components are acquisitions. Talk a little bit about just sort of how the mechanics of those conversations of the projects, of the people from the different BU’s get together. Was there already a really good solid foundation that then they could build on, or how did that work?
Uri Sarid: So, it’s interesting. The most important element of that foundation is a shared set of values. It’s actually not technology at all, it’s not interfaces, it’s are we actually trying to do the same thing, and do we believe that they’re important? When we looked at our values, it was values towards our stakeholders. That actually makes all the conversations a lot easier.
That’s actually not just a Salesforce thing. When we look at our customers, the ones that have been able to do things that normally would have taken them a year or two are now doing it in a matter of days. What actually allowed them to do that rapidly, it’s the fact that they had a common purpose. That common purpose may well have been, I’ve got thousands of employees I need to protect, or my business is going to go under if I don’t rearrange this, or I need to enable a certain something.
The degree to which I think as a society we’ve had a common purpose has allowed people to pull off things that were, in the past, almost psychologically blocked. Now it’s nice that you have a common purpose. Now do you have actually a way to actually turn that purpose into something real? And that’s where effectively, knowing what we’re good at and having places where we can attach to where other people are good at was really, really important. Right?
So as we just mentioned for the data platform, MuleSoft is good at particular things and we’re API interface and contract based, and we work really well with data models. So Salesforce has a data modeling team. They’re really good at working with those data models. It allowed us to actually each get good at what we’re doing at and align behind that common purpose.
There’s example after example of that. If you look at Work.com, a lot of pieces of Salesforce are actually relatively modular and extensible. And to that extent, taking the existing Salesforce platform and orienting it towards opening up was actually not a herculean task where everybody has to build a ton of software, but actually take your current capabilities around the ability to change schedules, for example, and schedule management. Or the ability to surface external data as we’re doing with a data platform. The ability to actually create portals. All of those capabilities were leveraged in very short order, literally in a matter of weeks, to bring in that new package to market.
Application networks are key to quickly adapting
Bill Detwiler: Yeah. I’d love to hear your thoughts in general on kind of that modularity and that extensibility question. Because I think when it comes to system design, platform software design, SOA has been around for a really long time. We talk about kind of modularity. You can get really super kind of granular when it comes to microservices and things like that. It seems like that that type of architecture, you almost need that type of modularity to provide that flexibility and the adaptability to handle events like the coronavirus. It’s the old monolithic design process just doesn’t seem as capable to handle that as the new models.
Uri Sarid: Yeah, I will be more emphatic than you. It is impossible to do it with a really large monolith. Right? The whole definition of a monolith is that all the dependencies are baked in. So if you have, say, a particular capability around, let’s say, new sales. If you need to repurpose that for something else, you just can’t. It’s built for one purpose, and all of the pieces that are stacked up behind it, how do I create a new organization, how do I surface the data and so on, are all pre-baked. There is no way to carve out a piece of it and say, “From this piece based just on its interface, I’m going to use it for something else.”
So, that modularity is absolutely key. But note, it is not the modularity of just sort of an IT concern. It’s not a computer network modularity. It’s actually an application network modularity. This is why we call it application networks. The application level, the capability and business level needs to be modular so you can plug and play those pieces. That’s where I think we’re starting to see application networks really play out. When people talk about agility, they always talk about, “Well, the new need of the consumer will be this and that. I’m going to position myself.”
No, the new need for the consumer yesterday was to be able to buy from you without entering the store. So how are you going to be agile instantly? How are you going to take your child protection workers and be able to deploy them into homes that may or may not have COVID-19? Those are real immediate needs for change, and only by having a modular application network can you actually do those. And we saw example after example of that.
Bill Detwiler: It seems like, at least to me, that those were changes that were happening before the current COVID-19 outbreak, but they were happening at a much slower pace and companies weren’t necessarily jumping to get them done rapidly. Some were. We’ve been talking about digital transformation for a really long time, but it seems like with the current pandemic, people finally realize, “Oh, now I have to do this.” It’s a matter either of, as you talked about, survival of my business or survival of my customers or survival of my employees. Now it’s an imperative that we do this, not an optional.
Uri Sarid: I think that’s absolutely right. We were already quite happy with the pace of innovation before this. We’ve talked about the new industrial transformation, the new industrial revolution. We had seen over and over again that the companies that take innovation and agility seriously are the ones that survive. We were talking about every two or three years you start to see really new innovations and survival of the fittest. We were pretty happy with the pace of innovation before this.
Relative to today, that still looks a little bit like the old boiling the frog, right? The temperature was getting hot, but some of the legacy companies were still like, “I think I can still pull this off for a few more years.” And then you say, “Well, how about a few more days?” The differentiation between those that can and those that cannot, well, the ones that didn’t have VPN and work from home literally could not operate. They had to stop all of their sources of revenue because they had to go and retool. The ones that had APIs and external connectivity and had the building blocks pivoted very quickly.
In fact, I think there’s something interesting that comes of this. At some point the pandemic will be behind us. At some point, hopefully sooner, but it might take another year to truly put this one behind us, the thing that will stay with us is people don’t easily forget this kind of speed. They will turn around to leaders and say, “Well, last year you managed to make that happen in three weeks. Why do you think that it’s going to take six months right now?” And that speed, I think to some extent will be with us for a long, long time.
Arguably, that’s also what’s going to help a lot in coming out of the economic downturn, which I think will likely outlast the pandemic by quite a while.
MuleSoft 2020 roadmap and beyond
Bill Detwiler: Speaking about the future, what’s up next for MuleSoft? What’s on the horizon for the work that you’re doing within Salesforce?
Uri Sarid: I think what you’ll see is … And again, fortunately we were already on that trajectory, but you’ll see us continue to double down on that, is the ability to enable everyone in a company and enable more and more companies to participate in building out application networks declaratively. So you will see us come up with an offering that actually embeds inside of Salesforce itself as a primary example of a SAS provider that allows the Salesforce users to directly create automations and integrations.
That same product, of course, will be offered to the general market the following year to integrate anything with anything, and that will show the world how you can get the same kind of acceleration that you saw for IT for anyone, and really the democratization.
The other aspect is really to continue to invest in the metadata driven or the declarative aspects of our platform. We’re already very declarative. What it means literally is that all you have to do is you have to capture the intent, I would like to expose the following capabilities, or I would like to connect the following sources of data, or I would like to impose the following security restrictions, and then you let the computing infrastructure take care of it. You let the platform of the application network actually take care of it. We will do more and more of that.
One of the cool features that you will see coming up later this year is something that’s called API Federation. What it means is that if you’ve already exposed the set of APIs and you’d like to consume data that’s exposed across multiple of them but you’d only like subsets, effectively like a slice and dice of the data or looking at your API as kind of a distributed database, can I just launch a query and just get the data out that I want to without having to create a new node in the application network? That capability will be coming out later this year and will I think materially improved developer productivity. So you’ll see us start to really leverage that investment in application networks to make developers go faster and to make non-developers empowered, as well.